1. Field of the Invention
The present invention pertains to a method and device to determine whether or not edges of the image of an original document read by an image sensor are black edges in an image processing device mounted in a color copying machine, etc., as well as to an image processing device using said identifying method and device.
2. Description of the Related Art
In conventional image processing devices, in order to improve the reproduction quality of black characters or black lines (line drawings) included in the color image of an original document read by the image sensor, the edges of the black characters or black lines (hereinafter collectively called ‘black characters’) in the image are generally identified, so that image processing such as edge enhancement may be performed based on the results of the identification. This image processing is briefly explained below.
First, the edges of the image are identified by filtering, using a primary differential filter and a secondary differential filter, the input image signals obtained through the reading of the original document, and if the saturation is lower than the threshold value, which is set in accordance with the brightness information of the input image signals, the image is identified as a black character. Based on these determination results, black edges (any area that comprises an edge and is part of a black character) are identified. Whether the area is outside the edges (the background side) or inside the edges (on the black line) may be determined based on whether the output of the secondary differential filter is positive or negative. In the explanations that follow, the outside area of a black edge will be called the black outer edge area, and the inside area of a black edge will be called the black interior edge area.
In order to improve the reproduction quality of black characters or black lines, the following processing is carried out regarding the black interior edge area and black outer edge area. For the pixels in a black interior edge area, the edge enhancement process is performed, wherein a brightness component is added to the black image data K. For color image data C (cyan), M (magenta) and Y (yellow), no edge enhancement is carried out, and processing is performed in which the image data for the target pixel is replaced with pixel data having the smallest value in a 5×5 or 3×3 pixel matrix (in other words, the pixel data having the lowest darkness level).
For pixels in black outer edge areas, edge enhancement is not performed to either the black or color image data K, C, M or Y, and processing is performed in which the image data for the target pixel is replaced with the pixel data having the smallest value in a 5×5 or 3×3 pixel matrix.
Using such processing, C, M and Y color components are reduced near the edges of a black character or black line in the image of an original document while the black interior edge areas are enhanced, resulting in improved reproducibility of black characters or black lines.
As described above, determination as to whether or not edges are black edges has conventionally been made based on a comparison of the saturation information with a threshold value obtained based on the brightness value. Where the saturation is smaller than the threshold value, the image may be deemed to be a black character. However, it is empirically known that the higher the brightness, the higher the required threshold values are.
In addition, a determination method disclosed in Japanese Laid-Open Patent Application No. Hei 8-116462 is also proposed in order to increase the accuracy of black character identification. In this method, a second black character identifying unit is used in addition to the identifying method described above (the first black character identifying unit). While the first black character identifying unit identifies black characters based on the brightness information and saturation information obtained from the brightness/chromaticity Lab space, the second black character identifying unit identifies black characters using the black signals (K) and saturation signals obtained from the YMC color space. A black character is identified only when both the first and second black character identifying units output a black character identification signal.
However, in the conventional method described above, when the balance among the three primary colors (R, G, B) is compromised due to aberrations in the optical system or the line-to-line correction, performing accurate black character identification becomes difficult.
In the identifying method disclosed in Japanese Laid-Open Patent Application No. Hei 8-116462, where color balance is compromised due to signal processing such as line-to-line correction, the identification accuracy will be improved. However, because the YMC color signals, which are the source of the black signals (K) and saturation signals input to the second black character identifying unit, are signals obtained by converting the Lab space signals, where the color balance, which comprises the Lab space signals themselves, is compromised due to aberrations in the optical system, no improvement in identification accuracy can be expected.